ommv0.2.148

omm install

Download a model into the central hub, link it into every installed runner, and check it fits this machine's memory before spending your bandwidth.

01 / 06

Overview

Reach for install once search or list has given you a name, a repo reference, or a numeric index. install checks the model against this machine's predicted memory budget before downloading anything, links the finished file into every runner installed on this system, and prints the exact commands to run or uninstall it. A name that resolves to more than one quantization or more than one provider drops into an interactive picker instead of guessing.

02 / 06

Options

Every flag this command accepts, and what it defaults to when you leave it out.

  • <name>—Default: required

    A curated name, a numeric index from the last search or list run, an 'org/repo:file.gguf' reference (optionally prefixed hf: or ms:), or a direct URL.

  • --skip-unfit—Default: off

    If this hardware is predicted not to run the model, skip it instead of asking. Exits 0 with skipped_unfit set — for scripting.

  • --upload / --no-upload—Default: current upload policy

    Send (or skip sending) this machine's benchmark result to the telemetry server without asking. Left unset, the current omm setting upload policy decides.

  • --force—Default: off

    Re-download even if this model is already installed.

  • --verify-runtime / --no-verify-runtime—Default: asks first

    Run (or skip) a short local load/generation check after linking. Left unset, install asks before loading a model that isn't already running.

  • --json—Default: off

    Accepted as a global flag, but install has no JSON output mode; omm prints a warning and ignores it.

  • --yes, -y—Default: off

    Skip confirmation prompts, including hardware-fit and runtime-load consent, for non-interactive installation.

  • --quiet, -q—Default: off

    Suppress progress bars and background status or hint lines. Errors and the final install result still print.

  • --no-color—Default: off

    Disable colored terminal output.

  • --help—Default: off

    Print install's usage, arguments and options, then exit.

03 / 06

Examples

From a plain search to something you'd put in a script.

Plain install — checks hardware fit, downloads, verifies, links.

$ omm install mistral-7b-instruct-q4

Skip this model instead of asking, if it's predicted not to run here.

$ omm install mistral-7b-instruct-q4 --skip-unfit

Install by the number the last search or list run printed.

$ omm install 1

Install without sending a benchmark result, regardless of the saved upload policy.

$ omm install mistral-7b-instruct-q4 --no-upload

Re-download even though this model is already installed.

$ omm install mistral-7b-instruct-q4 --force

04 / 06

Recorded CLI run

Command recorded in this video

omm install hf:example/oversized-405b-gguf:oversized-405b-q4_k_m.gguf --no-color

exit code: 0 · outcome: cancelled

Cancelled by the recorded keyboard input before any install, model run, benchmark, or upload.

Open the exact real-process transcript (.txt)

Extended documented terminal example

Format-accurate reproduction of the real download, checksum and link-summary lines (src/omm/downloader.py:107-170, src/omm/cli.py:4877-4886) — not a literal capture, since actually downloading a 4.4 GB file isn't something this page does. Filename, byte count and the three linked runners match the site's own verified install demo (design/FACTS.md).

06 / 06

If something goes wrong

Every message below is one this command actually prints. Find yours, read why it happened, then do the last line.

  1. Unknown model 'zzzz-totally-fake-model-name-xyz'. Use a curated name (tinyllama-1.1b-q4, llama3.1-8b-instruct-q4, mistral-7b-instruct-q4), an 'org/repo:file.gguf' ref (optionally prefixed 'hf:' or 'ms:'), or a direct URL.
    why
    This name doesn't match anything in the curated catalog, and it isn't a repo reference or URL install recognizes either.
    what to do
    Try omm search first to find the exact name or reference, or pass an 'org/repo:file.gguf' reference directly.
    source
    src/omm/hub.py:370-374
  2. Not enough disk space: /Users/you/.omm/models needs up to 5.2 GiB (central model download) but only 3.1 GiB is free.
    why
    install checks free space before downloading — the central model file, plus whatever a runner needs to copy or link it, plus a safety margin.
    what to do
    Free up space on the reported volume, or point OMM_HOME at a roomier drive before installing.
    source
    src/omm/cli.py:3479-3485

Still stuck? Open an issue with the exact message you saw.